Network analyzers are devices that are used to determine the radio frequency (RF) characteristics of various devices under test (DUTs). In many situations, a DUT is a relatively small component designed to interface with a trace contact point on a printed circuit board (PCB). Many network analyzers typically utilize an interface adapted to receive a coaxial coupling. To test a DUT designed to be employed on a PCB using such a network analyzer, a test fixture is often employed. A test fixture is generally a specialized device that is adapted to readily accept a DUT and that electrically couples the DUT to one or several ports of a network analyzer.
For many DUTs (such as balanced filters, baluns, balanced amplifiers, etc.), the pertinent performance measurements depend upon both the magnitude and phase of the signals applied to and received at each port. In the case of balanced devices, it is quite important that the loss of each port be identical between the balanced pairs of ports. However, the use of network analyzers and test fixtures to perform such measurements presents difficulties. Specifically, it is common to experience different path lengths on different ports using test fixture/network analyzer configurations. The variations may result from PCB layout constraints, manufacturing process limitations, and/or other reasons. The variations between ports may cause the loss to vary between the port thereby reducing the accuracy of the measurements of a DUT using the test fixture. The amplitude loss in the test fixture may cause a properly functioning part to fail insertion loss tests. The amplitude loss can also make matching measured results to model predictions difficult.
De-embedding an S-parameter description of the test fixture has been used to address loss in the text fixture. In known implementations, de-embedding requires a network analyzer to be operated in an error correction mode and involves multiple analyzer sweeps to generate measurement data to be corrected. The measurement data is then processed on an off-line basis by applying one or several error correction arrays using matrix operations. Also, known error correction arrays are statically defined in error correction files. Furthermore, the error correction arrays involve a one-to-one relationship between error terms and the spectral data. Accordingly, it is often the case that the requirements for de-embedding a test fixture are difficult and are often not practical in normal manufacturing environments.